Geometry and Architectural Planning at Göbekli Tepe, Turkey

Gil Haklay & Avi Gopher

The site of Göbekli Tepe in southeastern Turkey keeps fascinating archaeologists as it is being exposed. The excavation since 1995 has been accompanied by a lively discussion about the meaning and implications of its remarkable early Neolithic megalithic architecture, unprecedented in its monumentality, complexity and symbolic content. The building history and the chronological relations between the different structures (enclosures), however, remain in many ways a challenge and open to further analysis. The study presented here is an attempt to contribute in this direction by applying a preliminary architectural formal analysis in order to reconstruct aspects of the architectural design processes involved in the construction of the monumental enclosures. This is done under the premise that such investigation would shed light on the chaîne opératoire of the enclosures’ construction and their history, thus enabling a fresh look as well as an evaluation of past suggestions regarding these structures and the people who built them. Indeed, the results of the analysis brought to light an underlying geometric pattern which offers a new understanding of the assemblage of architectural remains indicating that three of the stone-built large enclosures were planned and initially built as a single project.

Introduction throughout the Levant and Cyprus, recapitulates the advances of architectural planning (thought, In the archaeological record, architectural planning principles and methods) during this time, as geomet- that includes speciﬁcations of architectural spatial ric architectural planning is often detectable while forms prior to construction becomes archaeologically analysing their remains. The site of Göbekli Tepe, visible with the appearance of stone-built shelters in however, stands out when it comes to the volume of the late Epipaleolithic period (Natuﬁan sites) in the construction endeavours undertaken, among other Levant (Haklay & Gopher 2015). Both the planning things, and, as we will show and discuss, the com- and the execution of the plan with accuracy were plexity of the planned and built architectural projects. made possible by conceptualizing basic geometric This study will discuss the building history of ideas and methods such as circle, centre and compass the monumental enclosures in the main area of arm. Subsequently, the early Neolithic period has Göbekli Tepe, as well as the chronological relations seen the emergence of substantial built environments between them, by applying a preliminary architec- in the form of villages and beyond. During the tural formal analysis that reconstructs aspects of the Neolithic period, from its very beginning, large con- architectural design processes involved in the construction projects were built, including the Jericho struction of these enclosures. We assume that such tower and other non-domestic, communal/collective analysis would shed light on the chaîne opératoire of structures. The archaeological record of these so the enclosures’ construction (mostly the planning, called ‘special buildings’, found in Neolithic sites laying out the plan on the ground—and see below,

Cambridge Archaeological Journal 30:2, 343–357 © McDonald Institute for Archaeological Research 2020. This is an Open Access article, distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives licence (http://creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is unaltered and is properly cited. The written permission of Cambridge University Press must be obtained for commercial re-use or in order to create a derivative work. doi:10.1017/S0959774319000660 Received 19 Mar 2019; Accepted 6 Oct 2019; Revised 6 Aug 2019 Downloaded from https://www.cambridge.org/core. IP address: 170.106.34.90, on 28 Sep 2021 at 20:02:34, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0959774319000660 Gil Haklay & Avi Gopher

This paper will focus on the architectural remains excavated in the southeastern part of the site (the main excavation area) under the direction of Klaus Schmidt in the years 1995–2014, on behalf of the German Archaeological Institute and the Museum of Şanlıurfa (O. Dietrich et al. 2014; Peters et al. 2014; Schmidt 2000; 2001; 2002a,b; 2003; 2004; 2006; 2007; 2008; 2010; 2012). The site is located at the centre of the Fertile Crescent, at the ‘heart’ of the Golden Triangle (Aurenche & Kozlowski 2005; Schmidt 2000) and in the Core Area of plant domestication (Abbo & Gopher 2017; Abbo et al. 2010; Gopher et al. 2017; Lev-Yadun et al. 2000). It was suggested by the excavator that early layers of the site were built and used by hunter-gatherers, since no domes- ticated plants or animals have been recovered. The stratigraphy of the site comprises two main layers. The older Layer III assigned to the Pre-Pottery Neolithic A (PPNA) period is characterized by large curvilinear enclosures, while the younger Layer II assigned to the early and middle Pre-Pottery Neolithic B (PPNB) periods features relatively smaller, rectangular structures, typically with lime-plaster floors (Schmidt 2002a) and crowded together with shared walls. A third, maybe intermediate layer, marked as ‘uncertain’ on the schematic plan (Fig. 1; O. Dietrich et al. 2014, fig. 2), contains small free- standing structures of irregular shape contours, Figure 1. The main excavation area: schematic plan. defined both by curvilinear and straight wall seg- fi (Modi ed from O. Dietrich et al. 2014.) ments. Based on a geomagnetic survey, it is esti- mated that at least 15 more Layer III curvilinear enclosures are scattered throughout the mound including Figure 8—and initial construction stages) (Schmidt 2003). According to the excavator, the thus enabling a new understanding of the assem- enclosures of Layer III went through a series of back- blage of these architectural remains. We first present filling events, interpreted as an intentional ‘burial’ of the site and the enclosures of the main area of these round monumental enclosures (Schmidt 2000). Göbekli Tepe, then our methodology and the archi- These enclosures are the subject of our analysis here. tectural formal analysis, and finally the results of Four Layer III enclosures (Enclosures A–D) were the analysis. The discussion closing this paper con- uncovered so far in the main excavation area, each cerns questions of architectural planning and the sig- featuring a number of often decorated monolithic nificance of our results vis-à-vis earlier interpretations T-shaped stone pillars (3–4m high)embeddedinto of the architectural ensemble of the main area at stepped-back peripheral walls and set around a pair Göbekli Tepe. of central larger stone pillars (up to 5.5 m high). The lower part of the peripheral walls (the ‘bench’) typic- Göbekli Tepe ally interconnects the inner edges of the T-shaped pillars. In their final form, structures B, C and D The Pre-Pottery Neolithic site of Göbekli Tepe is were curvilinear enclosures (10–30 m in diameter), pos- located at the summit of a limestone mountain sibly partially subterranean and roofed (L. Dietrich ridge in the Şanlıurfa Province, southeast Turkey. It et al. 2019 and references therein), each defined by is a 15 m high artificial mound covering an area of one or more circumventing peripheral walls (which about 9 ha. Excavations carried out in different may also represent different construction episodes). areas of the site yielded megalithic architecture Enclosure A (which based on radiocarbon dating is 1 dated to the twelfth and eleventh millennia cal. BP assumed to be the most recent one) is characterized (tenth and ninth millennia cal. BC: Schmidt 2010). by elongated walls. It has been suggested that

344

Downloaded from https://www.cambridge.org/core. IP address: 170.106.34.90, on 28 Sep 2021 at 20:02:34, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0959774319000660 Geometry and Architectural Planning at Göbekli Tepe, Turkey

Enclosure A may represent a transitional phase from peripheral wall (about 9 m in diameter) surrounding round to rectangular architecture (O. Dietrich 2017a; the two central pillars includes nine T-shaped pillars O. Dietrich et al. 2013). so far. The top of walls exposed north and south of One of the most outstanding characteristics of the peripheral wall may suggest a second concentric the site is the richness in imagery items (especially) wall (about 14 m in diameter), so far without pillars. associated with the megalithic architecture. Many of The enclosure seems to have had a lime-plastered the architectural elements including pillars, pillar floor, which was exposed mainly between the two bases, port-holes and gates bear representations in central pillars. By the eastern central pillar, a stone relief, depicting a wide range of wild animals. slab with a shallow channel leading to a carved Mostly, the monolithic T-shaped pillars bear represen- bowl embedded in the floor hints, according to the tations in bas-relief, but high-relief representations excavator, that a ritual may have taken place in the and smaller 3D sculptures integrated in the walls space between the twin pillars (O. Dietrich et al. have been found as well (Schmidt 2006; 2007; 2008; 2012a; Schmidt 2001), both bearing bas-reliefs of a 2010; 2012). Of the Layer III T-shaped stone pillars, south-facing fox on their inner side. According to the two central pillars of enclosure D stand out, as the research team, the lime-plaster floor may cover they depict anthropomorphic ‘beings’,withthehead an older bedrock floor similar to the floors in represented by the traverse top of the T-shape, while Enclosures C and D (O. Dietrich 2017b). The setting the body is represented by the vertical part of the of a number of the T-shaped peripheral pillars bear- pillar (‘shaft’) on which orthogonal projections of ing reliefs on their narrow rear side (away from the arms, hands, belt and a possible so-called groin centre) (Peters & Schmidt 2004) may indicate that cloth are depicted on three sides (Pillars 18, 31). they are re-used. A rectangular port-hole stone was Similar anthropomorphic T-shaped pillars, although found in the backfill material at the centre of the smaller, were first excavated in the cultic terrazzo enclosure, south of the central pillars. Another port- building of the Early Pre-Pottery Neolithic B hole, found in situ embedded in a wall (O. Dietrich (EPPNB) site of Nevali Çori (Hauptmann 1993). In 2017b) north of the first peripheral wall, seems to Göbekli Tepe, too, the T-shaped pillar tradition con- align with the axis of symmetry that runs between tinued into the PPNB although the T-shaped pillars the central pillars of this enclosure. featuring in the PPNB rectangular structures (Layer II) are much smaller (about 1.5 m high, similar to Enclosure C the Nevali Çori pillars) and less frequently decorated. The remains of Enclosure C comprise at least two Small-sized T-shaped monoliths are also known from concentric peripheral walls (possibly three, according several Pre-Pottery Neolithic (PPN) sites in the to Piesker 2014). The inner wall (about 11 m max- Şanlıurfa region, around Göbekli Tepe (e.g. Sefer imum interior diameter) has been completely Tepe, Karahan Tepe and Hamzan Tepe) (Çelik 2010; exposed. It includes a peripheral bench-like inner 2011;Güleret al. 2013). However, while the depictions lower part which connects the edges of the 11 periph- carved on the pillars seem to have been shared across eral T-shaped pillars. The ‘bench’ follows a continu- a wide territory, the pillars themselves (the platform) ous oval curve which is disrupted between the two were more limited in distribution. For example, the southern pillars forming a niche or an entrance that snake with a triangular head is a common motif was successively blocked. The ‘bench’ (up to 1.6 m both in Göbekli Tepe and PPNA Jerf el Ahmar high above the bedrock floor) was not found capped (Helmer et al. 2004). with stone slabs (as for example in Enclosure F, west Other than the megaliths, construction technol- of the main excavation area). Instead, stone slabs ogy included rock-hewn and lime-plaster (‘terrazzo’) (possibly recycled T-shaped pillars) were found in floors, walls built of dressed stones (sometimes front and to the sides of the southern peripheral worked on all sides) in both the external and internal pillars, resting on additional bench-like or platform faces with a core of earth and smaller stones in features running parallel to the long side of the between, and the use of mud mortar and wall plas- rectangular space between the central pillars. The tering (preserved in Enclosure D). floor of the enclosure is a carefully levelled bedrock surface which includes two rock-hewn platform- A brief description of the enclosures like pillar bases for the ‘twin’ central pillars. Interestingly, the pillars are not centred on their Enclosure B bases but set closer to the peripheral walls, leaving Enclosure B was not fully excavated. Its western a larger platform surfaces towards the centre part is still superimposed by PPNB structures. The (Fig. 2). That part of the eastern pillar base features

345

Figure 2. Enclosure C. Note the discontinuity of the bench between the southern pillars and the additional construction built against the southern part of the inner peripheral wall. (Reproduced from O. Dietrich et al. 2012b.)

two circular depressions, and on top of it, a sculpture southern part, the second peripheral wall connects of a boar and ground stone implements have been to an elongated passageway (‘dromos’) which leads found (O. Dietrich et al. 2012b, fig. 6; Schmidt from a gate complex that includes a port-hole stone 2008). The peripheral T-shaped pillars of the inner and a decorated monolithic U-stone (Schmidt 2012, peripheral wall (sometimes broken and mostly 124, 148–9). According to the excavator, sometime re-used) were integrated in built ‘pockets’ at different in antiquity, a large robber pit (10 m diameter and elevations within the wall to achieve a common top 3 m deep) had been dug at the centre of the enclosure elevation. According to Piesker (2014), the construc- penetrating the backfill, in what looks like a deliber- tion seams, usually placed on both sides of the per- ate action against the buried central pillars, whose ipheral T-shaped pillars, indicate that the inner broken parts were found on the pit’s floor (Schmidt peripheral wall was built in its entirety in a single 2008). The western central pillar bears a bas-relief construction episode, leaving sockets in the wall for of a fox. The inner face of the eastern pillar was the later installation of the pillars (with the exception badly damaged, but according to the excavator the of the blocking of the southern entrance and the remains of a bull depiction are still visible. installation pillar P40 that was identified as a later Enclosure C is rich in decorated pillars: most notable addition). The second peripheral wall (with an is Pillar 27 that bears a high relief of a predator on its inner diameter of about 17 m) features eight pillars narrow inner face (Schmidt 2006). so far. Its northeastern part has only been partially excavated. In the southwestern part of the structure, Enclosure D a lime-plaster floor was preserved in the passageway Enclosure D has been described as the largest and between the two peripheral walls at an elevation of best preserved of the Layer III enclosures in the about 2.1 m above the central bedrock floor. At its main excavation area (e.g. O. Dietrich et al. 2012a;

346

Notroff et al. 2014; Schmidt 2012). It also has the most et al. 2018; O. Dietrich et al. 2013; Piesker 2014) which symmetrical (oval shape) contour and pillar arrange- remain a primary topic of investigation (Notroff et al. ment. The maximum inner diameter of this enclosure 2014). It has been suggested that remodelling phases is about 14 m. Its peripheral wall, preserved to an ele- consisted of building a peripheral wall of smaller vation of over 3 m, connects (so far) 11 T-shaped pil- diameter, which was set inside larger, earlier ver- lars, some of which are decorated with complex sions of the enclosures (e.g. L. Dietrich et al. 2019; depictions. Its central pillars with anthropomorphic Hodder in Banning 2011; Schmidt 2012; Watkins depictions, rising to a height of 5.5 m, are the largest 2004).2 It has also been suggested that at the end of megalithic features on the site. As in Enclosure C, the the use life of each enclosure, it was filled and a twin central pillars were set on shallow pedestals or new enclosure was built alongside (Watkins 2010), platform-like bases (20–30 cm high) carved out of or more specifically, that based on somewhat the bedrock forming the enclosure floor. The southern younger 14C dates, the outer peripheral wall of face of the eastern pillar base is decorated with bas- Enclosure C may have been built during the backfill- relief. The twin central pillars are facing south, and ing of Enclosure D (O. Dietrich et al. 2013). According one of them (the eastern) has a bas-relief of a fox on to the excavator, this remained an open question, its inner side (as in Enclosures B and C), depicted admitting that it is possible that the Layer III enclo- above the arm as if it is being carried by the anthro- sures date from the same time, while it is also pomorphic being (Schmidt 2007). In the southern possible that they were built in succession (Schmidt part of the enclosure, the remains of a second periph- 2012, 215). eral wall have been exposed, although this wall is The following architectural formal analysis is built differently than the outer peripheral wall of an attempt to shed some light on the chronological Enclosure C, as it is more segmented and includes relations between and within the enclosures by a protruding straight walls which connect to the nar- reconstruction of certain aspects of their initial row rear side of the peripheral pillars, dividing the design process, which further nuances the chaîne space between the concentric walls into cell-like opératoire of their construction, especially the enclosures. It has been suggested that those radial early stages of planning and laying out the wall segments were meant to counter the thrust of plan on the ground. This will be achieved by roof beams (Banning 2011). identifying the spatial principles and compositional laws governing the generation of the structures’ Enclosure A form. Enclosure A is only partially excavated, and although its floor has not yet been reached, its floor Methodology plan already seems different than the other Layer Architectural formal analysis, which studies the III enclosures. Unlike the radial arrangement of the spatial forms of architectural built spaces, is used to peripheral pillars in Enclosures B–D, the four periph- trace back aspects of architectural planning pro- eral T-shaped pillars exposed so far in Enclosure A cesses, construction and performance. Such analysis are all oriented parallel or perpendicular to the direc- may indicate, for example, the use of a pre-planned tion of the central pillars. The central pillars are not schema and its adaptation to a specific site, or free-standing as in the other enclosures, but attached highlight the internal logic that guided the architec- to straight wall segments that extend towards the tural design process. In order to detect spatial south and connect by an apse-like curved wall on relations that may not be noticeable at first glance, their northern side. According to the Göbekli Tepe we use an analytic tool (an algorithm) based on Research Team, the eastern central pillar (P2) does standard deviation mapping to study spatial form not stand in its original position, since in addition and relative location of architectural features in to other field observations, it bears reliefs on its space. The algorithm, which we have previously narrow, northern side. Radiocarbon dates have con- used to analyse Natufian structures (Haklay & firmed the excavator’s hypothesis that this enclosure Gopher 2015), produces a statistical centre point is younger than Enclosure D (O. Dietrich et al. 2013, based on the relative position of architectural ele- and see note 1). ments. In a second phase of the analysis, the math- ematically identified centre point is examined in Architectural formal analysis relation to the whole assemblage of architectural remains (other non-tested features), in order to find So far, very little has been written about the construc- out if it may suggest a geometric regularity or indi- tion history of the Layer III enclosures (but see Clare cate a point of significance.

347

Results

Analysis—phase A We first examined Enclosures B, C and D that share a basic layout which can be described in the most abstract way as a pair of central pillars, surrounded by peripheral pillars embedded in a curved wall. In order to articulate this common layout further, we searched for a regularity in the position of the central pillars within the enclosures. We used standard deviation mapping to detect statistical centre points within the enclosures, based on the relative (and final) location of the peripheral pillars embedded in the inner ring walls. Each pillar was represented by a point marking its approximate centre (these points best represent the location of the pillars regardless of their orientation), and in each enclosure, the centre point identified was the point at which the standard deviation value of the group of distances (from that point) to the points marking the peripheral pillars was minimal (Fig. 3). Figure 3. Architectural formal analysis: Phase The centre points in Enclosures C and D were A. Identification of a geometric and statistical centre point found to be at the middle point of the southern based on the locations of the peripheral pillars: edge of the quadrilateral space between the central visualization of the calculation. Note the red dots marking pillars (aligned with the front of the south-facing the peripheral pillars and the resulting centre points. anthropomorphic T-shaped pillars), while in (Drawing superimposed over the schematic plan. Modified Enclosure B it was near the centre of that space. from O. Dietrich et al. 2014.) These results suggest that there is an exact geometric relation between the central and the peripheral pillars. The standard deviation mapping method con- It is of note that the central pillars in Enclosures sists of a simple geometric and statistical calculation C and D were set into pedestal-like bases carved in carried out by an algorithm. The input for the algo- bedrock, which most probably pertains to the initial rithm is co-ordinates (relative to an arbitrary origin) construction of the enclosures, while the central pil- of physical points in space. For example, the points lars of Enclosure B are set in a plaster floor. Unlike may mark the centres of post-holes or trace the the central pillars, the peripheral pillars and walls path of a wall. A grid (e.g. 5 cm spacing) is first may have been subjected to modifications during superimposed over a plan of the architectural the life history of the enclosures. Yet each time a per- remains, and the algorithm proceeds by measuring ipheral feature was altered or added, it was neverthe- the group of distances from each grid point to the less carried out with reference (and full awareness) in set of input co-ordinates. Mean and standard devi- the mind of the builders to a point in space that ation values are calculated for each group. The out- marked the centre of attention in each enclosure— put is a statistical centre point in which the group the centre point. Although the exact construction of distances to the given input points has the lowest sequences of the enclosures are mostly unknown, standard deviation value. Thus, relative to that centre the same geometric rules seem to have been refer- point, the dispersion of the group of distances is min- enced in both the original construction and subse- imal (it equals zero if the points all lie on a circumfer- quent alteration episodes. ence of a circle). In the case of Enclosure C, whose central floor Since the excavation is on-going and an updated was completely exposed, we were able to verify the detailed plan has not yet been published, we have identification of the centre point also with respect used the most up-to-date schematic plan available to the curvature of the peripheral wall. We marked which proved sufficient for the purpose of our ana- on the detailed plan 32 evenly spaced points (at inter- lysis. In addition, we used the detailed plan of vals of 1 m) along the curve describing the inner face Enclosure C following Piesker (2014). of the peripheral wall (continuing the curve over the

348

Figure 4. Architectural formal analysis: Phase A. (Top) Identiﬁcation of a geometric and statistic centre point based on the curvature of the inner peripheral wall: visualization of the calculation. Note the red dots marking 32 points along the inner face of the wall and the resulting centre point. (Bottom) The central pillars and peripheral wall are equidistant from each other along the main axis. (Drawing superimposed over the detailed plan. Redrawn from Piesker 2014, ﬁg. 8.)

southern niche/opening where the bench is missing Further veriﬁcation that the central pillars are to achieve a continuous contour) (Fig 4). We applied precisely positioned within the enclosures (as in the standard deviation mapping algorithm with rela- Enclosures C and D) is provided by Enclosure F, tion to these 32 points and veriﬁed that the centre west of the main excavation area. We used the point indeed results exactly at the midpoint between same standard deviation mapping (algorithm) to the southern faces of the central pillars. In addition, detect a statistical centre point which similarly we noticed that along that axis the distance between resulted in a point between the central pillars, the centre point and the inner faces of the central pil- aligned with their southwestern extremities lars (about 2.8 m) is equal to the distances from the (Fig. 5a). Finally, it seems that the same abstract inner faces of the central pillars to the inner face of rule was followed also in the PPNB cult building at the peripheral wall (Fig. 4). Nevali Çori. There, a pair of anthropomorphic

349

Figure 5. (a) Göbekli Tepe Enclosure F (modiﬁed from O. Dietrich et al. 2014); (b) Nevali Çori.

T-shaped pillars was set within a rectangular floor ropes of that same length pegged in the first two points defined by peripheral pillars and bench, and, as (thisisfurtherdiscussedbelow;andseeFigure 8). was the case at Göbekli Tepe, it is the edge, not the centre, of the central pillars that aligns with the Analysis—phase B main axis of the structure (Fig. 5b). The example of We further analysed the remains of the three enclo- Nevali Çori suggests that, along with the tradition sures in order to find out if the possible underlying of having two central anthropomorphic T-shaped pattern may refer to additional properties of these pillars, the design rule regarding their positioning structures and other features on site. We noticed may have persisted as well. that the equilateral triangular pattern highlights the Once this design rule that regulated the posi- alignment of the central pillars of Enclosures B and tioning of T-shaped pillars at Göbekli Tepe Layer C, along an axis parallel to its southern side III was identified, we focused our attention on a spe- (Fig. 7). Furthermore, the composition and orienta- cific location in each enclosure, i.e. the middle point tion of the central pillars implies an axis of sym- between the front narrow sides of the central pillars metry. We found that the two Layer III U-stone —a point in space that has the spatial role of geomet- gates marked on the schematic plan are symmetric- rically tying together the central and the surrounding ally located and oriented on both sides of that axis peripheral pillars. In the main excavation area in (Fig. 7), which may further support this claim Göbekli Tepe, the spatial role of these points may (although the western U-stone’s relation to other have been extended beyond the boundaries of the architectural features has not been clarified). The single enclosures, as these points in Enclosures B, C eastern U-stone is found at the end of an elongated and D form an almost perfect geometric pattern passageway which connects on its other side to the too, being only 25–28 cm distant from the vertices second peripheral wall of Enclosure C. The passage- of an ideal equilateral triangle with a side length of way was built of large stones shaped on all sides about 19.25 m (Fig. 6). This precision with a distor- (O. Dietrich et al. 2014), and along with the 3×3 m tion of less than 1.5 per cent is less likely to be coin- U-stone monolith, they constitute a massive con- cidental, nor the result of a self-organization process. struction, which is likely to have been included in This basic geometric shape (the equilateral triangle) the initial (planning and) construction of the enclos- could have been measured simply by stretching a ure. It is possible that a similar passageway that led rope of the desired triangle side length to mark the from the western U-stone to Enclosure B was can- first two vertices of the triangle and then finding the celled by the successive (see note 1) construction of third vertex in the meeting point of two stretched Enclosure A.

350

Figure 6. Architectural formal analysis: Phase A. (In Figure 7. Architectural formal analysis: Phase B. (In red) red) The triangle defined by the three middle points The nearly equilateral triangle that passes through the between the southern faces of the central pillars; (In blue) middle points between the southern face of the central An equilateral triangle. (Drawing superimposed over the pillars of Enclosures B–D. (In yellow) The alignment of schematic plan. Modified from O. Dietrich et al. 2014.) the central pillars of Enclosures B and C along the southern triangle side. (In blue) The main axis, perpendicular to the southern triangle side, passes The axis of symmetry also implies a certain through the centre of Enclosure D. (In green) The fi architectural hierarchy, suggesting a special signi - U-stones symmetrically positioned on both sides of the cance to Enclosure D. This hierarchal organization main axis. (Drawing superimposed over the schematic fi of space is not only con rmed by the size of plan. Modified from O. Dietrich et al. 2014.) Enclosure D and the height of its central pillars, but most possibly also by the fact that the central pillars of Enclosure D are the ones bearing anthropo- simultaneously, but this possibility raises the ques- morphic attributes. The fox depiction that decorates tion of whether the initial construction of all three the central pillars of Enclosures B and C appears enclosures was more permeable, allowing lines of also on a central pillar of Enclosure D, but there it sight between their centres. As noted by Piesker is accompanied and dwarfed by the anthropo- (2014) and Clare et al. (2018), the peripheral morphic depictions. Enclosure D has also been T-shaped pillars embedded in the inner stepped per- singled out by the excavator as the only enclosure ipheral wall of Enclosure C (which may represent a that includes depictions of all the main game animals later construction phase) were never free-standing, (gazelle, Asiatic wild ass, wild sheep and aurochs) since they rest on the peripheral wall at different ele- and thus correlates with the faunal assemblage vations and were mostly installed in built pockets in (Peters & Schmidt 2004, 209). We therefore suggest the wall. However, less is known about the pillars of that, despite the numerous construction episodes the outer peripheral wall of this enclosure. Piesker and the continuous remodelling of the enclosures, it suggested that, due to the large size of the southern can be stated that the three enclosures B, C and D pillars (P11, P12, P23, PMA), they most probably were planned and initially built as a complex. This rest on bedrock, and that the northern P25 may be does not necessarily mean that they were built installed in a rock-hewn pedestal base similar to

351

Figure 8. (Top) The laying-out of an equilateral triangle. The equilateral triangle could have been laid out by following a simple method: 1. Marking the ﬁrst two vertices of the equilateral triangle (points A and B) by stretching a cord of the desired triangle side; 2. Pegging two cords of the same length (the desired triangle side length) at points A and B; 3. Stretching the cords until their other ends meet at the point marking the third equilateral triangle vertex. (Bottom) Geometric construction of an equilateral triangle.

the central pillars, so the issue of sight lines between our opinion, these findings provide strong support to the centres of the enclosures remains a possibility to our conclusion that Enclosures B–D originated as a consider. single project. In summary, we used standard deviation mapping to detect statistical centre points within the Discussion round Enclosures B–D, based on the relative (and final) locations of the peripheral pillars embedded The implications of the above analysis and its results in their inner peripheral walls. We found that these concern a number of issues and interpretations dis- centre points coincide in enclosures C and D with cussed by Göbekli Tepe investigators over the years the middle points between the southern narrow to which our results may add some insights. sides of the central pillars. This seems to be a design rule that determined the exact location of the central pillars within the enclosures. The identification of 1: A possible hierarchy between and within the enclosures this general pre-planning rule was confirmed by Our analysis revealed a clear hierarchically struc- the examination the remains of Enclosure F in tured space. The suggestions we made here, that Göbekli Tepe and the remains of the PPNB cult Enclosures B, C and D are one complex, highlight building at Nevali Çori, in which, similarly, the mid- the higher hierarchical position of Enclosure D, dle point between the front faces of the twin central which includes the axis of symmetry of the complex pillars is set in accordance with the (geometric and as a whole. Hierarchy is also implied at the statistic) centres of the floor in which they are intra-enclosure level considering the fact that the per- installed. We further noticed that these three points ipheral pillars are arranged around a centre point in Enclosures B–D at Göbekli Tepe form a rather between the fronts of the two central pillars—a exact equilateral triangle. The geometric pattern design rule that defines a precise spatial hierarchical was confirmed when examined along with the orien- relationship between central and peripheral elements tation of the central pillars (not yet taken into account (pillars). We may add the fact that the large central in our analysis), as the central pillars of Enclosures B pillars of Enclosure D, at the top of the hierarchical and C align along an axis marked by the southern order, are singled out as a unique class in the B–D side of the equilateral triangle. Further confirmation enclosures complex as a whole by their anthropo- is possibly offered by the symmetric position of the morphic depictions and prominent appearance. The in situ U-stones on both sides of the main axis (per- overlying symbolic layer, that is the depictions on pendicular to the southern edge of the equilateral tri- the pillars, further refines the hierarchy represented angle passing through the centre of Enclosure D). In by the architectural design.

352

2: The level of organization and manpower required ropes were used to measure the straight outlines on The level of organization and manpower required for the T-shaped pillars (Bar-Yosef 2014). It is possible the construction of the megalithic architecture of that ropes may also have been used to position the Göbekli Tepe (e.g. Bar-Yosef 2014; Notroff et al. central pillars of Enclosures B–D according to an 2014) should be multiplied by three, compared to architectural plan. The discerned geometric regular- previous estimations, if our suggestion is simply ities attest to an understanding of certain geometric accepted, since the potential size of a single project principles and their creative employment in the at Göbekli Tepe comprised three enclosures in the architectural design process. While the actual meas- case presented here. Even if this suggestion is uring of an equilateral triangle may not in itself be reserved, and beyond planning only the initial stages a difficult task, it represents an attempt to produce of construction were contemporaneous, this further a shape accurately by identifying intersection points amplifies the factors of manpower, organizational of curves traced by compass-arm techniques, which aspects, pace of construction and more. is an archaic form of a (‘Euclidian’) geometric construction (Fig. 8). The use of geometric construction in archi- 3: Was each enclosure constructed independently or not? tectural planning enabled the planner to conceptualize The suggestions based on the assumption that the a proportional abstraction of a floor plan of a rather structures were built and functioned independently, complex design, and to reproduce it at any size. either as residential units as suggested by Banning While the relatively simple architectural plans concep- (2011), or as public structures used by different (pos- tualized in the preceding late Epipaleolithic Natufian sibly competing) social groups as suggested by may have been mental ‘cognitive plans’ (Haklay & Schmidt (2000) and others (Peters & Schmidt 2004; Gopher 2015), during the PPNA, the dramatic Notroff et al. 2014), seem less likely (at least in the increase in complexity of the architectural design planning phase as well as at the beginning of their must have required the formulation of a schematic lives) following the results presented in this paper. (diagrammatic) small-scale floor plan which con- The suggestion regarding the different identities of sisted of a pattern, geometrically constructed and the builders of each enclosure, which accords well regulated by a length module (and see Haklay & with Schmidt’s early impression of the site as a ritual Gopher 2019 for a measure of unit used at Çayönü centre that drew people from as far as Jerf el Ahmar in the PPNB). The concept of a floor plan as an exter- and Nemrik (Schmidt 2002b), was further supported nal planning device is probably the biggest step for- by the identification of dominant animal species ward in Neolithic architectural planning. depictions in each enclosure, and their interpretation as emblems of different groups (although as long as 4: The nature of the T-shaped pillars the central pillars are considered, the depictions may A detailed discussion on interpreting the Göbekli express a linkage rather than differentiation between Tepe T-shaped pillars is beyond our scope here. We the enclosures, as the symbol of the fox appears in will concentrate on a few aspects regarding hierarchy every pair). Schmidt rightly noted that the depicted and totemism. reliefs are reproduced in the different enclosures It has been suggested in early interpretations of with an accuracy that could only be the work of a the Göbekli Tepe scene that the central T-shaped pillars school of specialists, and attempted to settle this of Enclosure D represent anthropomorphic ‘beings’ fact by suggesting that the enclosures may have such as ancestors (e.g. Peters & Schmidt 2004, 215). A only been ‘ordered’ by the different social groups similar suggestion was made for the other enclosures (Schmidt 2008). Our suggestion for a new under- and it was also extended to the peripheral T-shaped standing of the architectural remains of Enclosures pillars. The entity represented by a pillar bearing ani- B–D as constituting a complex (initially planned as mal depictions is therefore a combination of a human a single project) makes it less likely that the different (maybe an ancestor) and one or more animals. enclosures were initially built or commissioned by Simulating possible interpretations for the T-shaped different peoples/groups (Notroff et al. 2014; Peters pillars and the animals depicted on them, Peters and & Schmidt 2004), and it does not support the sugges- Schmidt (2004) discussed totemism as a possible inter- tion that one enclosure was being built while another pretation for the animal depictions. In this framework was being (or already was) covered up. specific relationships between man and animals are No less important, and central to this paper, is central. For example, the totem is often seen as a pro- the issue of architectural planning and the practical tector and in return the totemic animal is protected stage of laying out a construction project on the by a taboo forbidding its hunting.3 Peters and ground. It has been suggested that wild flax fibre Schmidt (2004) suggested that the pillars at Göbekli

353

Tepe could also be interpreted as poles linking the interpreted as reflecting a discourse related to dynam- underworld with the ‘living’ world. Following this ics of change in social order. We may speculate here line of thought, the T-shaped pillars may reflect both and say that this hierarchy represents a new world totemic entities and ancestors, as if the spirit of dead order, or the beginnings of social differentiation ancestors may operate in the world of the living reflected in the dead ancestors (or a dead ancestor) through totem animals. It is possible that the identity that are seen in the pillar itself and/or the depictions of the T-shaped pillars may have been conveyed both on it. This may have had significant repercussions in by their 3D form as ancestors and by the 2D depictions both the struggle over property (heredity order) and, they bear as animal forms that the ancestors may take most importantly, as the setting (incubation) of future in the world of the living. Considering our suggestion domestications as seen in the redefined relationships vis à vis the hierarchy of the enclosures and relating it to between man and certain animals. the depictions on pillars, we noted two elements: The discerned architectural hierarchy of the enclosures and the T-shaped pillars suggests that 1. The depictions of the central pillars of Enclosure D (the highest hierarchy) are unique in their the people of Gobekli Tepe were well acquainted clear anthropomorphic aspect (depicted on the with the concept of hierarchy. This accords well with statements made by Özdoğan (1999; 2014 and entire pillar shaft on three sides), and in size, which is by no means the largest. references therein) and by Schmidt (2012, 159) envis- 2. Fox depictions appear in every pair of central pil- aging a rich, complex hunter-gatherer social system possibly characterized by dynamics of a growing lars. In Enclosures B and C, a single fox is inequality during the PPNA in the northern Levant. depicted, and in Enclosure D it appears on the fi inner face of the central pillar at the height of Benz and Bauer (2013, 13, g. 2) also view PPNA social developments as a result of a rise in inequality the elbow of the depicted human arm. Schmidt suggested (2010, 244) that the figure ‘holds a fox and a decrease in sharing. They suggest that it was in the crook of its elbow’. In addition, following possibly single human agents (shamans, for example) that took part in creating the new restless Schmidt (2010) a fox-pelt was recognized (by its state of affairs and took advantage of it to ‘take long tail) in the depicted loincloth worn by the ’ anthropomorphic figures. Depictions of foxes on over the ritual arena and socio-political power. This would work well with our interpretation peripheral pillars are rare in Enclosures C and B and are common in Enclosure D. above that the identity of the T-shaped pillars (of Enclosure D) may reflect a (newly introduced) con- The first observation confirms the discerned spatial cept of human-to-human reincarnation rather than hierarchy of the D, C, B complex, while the second totemic human–animal relationships. observation suggests that the fox must have been of 3 greater importance than other totemic animals, but Conclusion its status in Enclosure D seems to have been different. Under this interpretation of the T-shaped pillars Architectural formal analysis (and the central points as totemic ancestral beings, in Enclosures B and C (of calculated) has brought to light an underlying geomet- the lower hierarchy) the fox (featuring on the central ric pattern based on an equilateral triangle and a set of pillars of both enclosures) was the main totemic ani- main perpendicular axes that ties together Enclosures mal (and see note 3). Enclosure D, however, is differ- B, C and D under a single, rather complex geometric ent in two respects: first, the depictions on the central design. This suggests a new understanding of what pillars suggest that the ancestors may take (in the has been initially planned and then built in the enclo- world of the living) a human rather than a non-human sure’s system of the main excavation area. It offers an animal form (which may be viewed as a sort of answer to questions on the chronological relationships reincarnation); and second, the depictions on the per- between the three enclosures, and it evokes insights ipheral pillars (in this enclosure only) include game regarding the architectural design process and how animals that comprised most of the basic meat con- such architectural complexity could have been achieved. sumed by the occupants (gazelle, Asiatic wild ass, Together with sites such as the PPNA Jericho in wild sheep and aurochs) (Peters & Schmidt 2004), the southern Levant with its complex construction and may thus reflect a different relationship between projects, Göbekli Tepe reflects a ‘foreign country’— humans and animals. Considering that the spatial a quantic jump of the latest hunters-gatherers in the arrangement of the structures, as we suggested, places region, a ‘wild growth’, a disorder of sorts in the Enclosure D at a higher order with respect to hunting-gathering world, way beyond the capacity Enclosures C and B, the complex as a whole can be of a pristine hunter-gatherer society. This has

354

eventually crystallized into new world-views, mainly in man–world and man–man relations that were sub- Gil Haklay sequently translated into new lifeways and a new cul- Sonia and Marco Nadler Institute of Archaeology tural landscape. The above attempt to incorporate our Tel Aviv University understanding of the outstanding architectural scene POB 39040 of Göbekli Tepe is only a beginning. The multitude Ramat Aviv of details that need analysis and understanding is Tel Aviv 69978 still overwhelming, but we are sure that interpreting Israel this site will continue and take new roads as the data- & base grows and its analysis continues. Israel Antiquities Authority POB 586 Notes Jerusalem 91004 Israel ‘ 1. Radiocarbon dating of the Layer III enclosures: The E-mail: [email protected] date of enclosure C (9261–9139 cal. BC. Charcoal. Sample recovered from between the outer peripheral Avi Gopher walls close to the bedrock) is younger than that of enclosure D (9745–9314 cal. BC. Clay mortar. Sample Sonia and Marco Nadler Institute of Archaeology recovered from the wall plaster between pillars 41 and Tel Aviv University 42). The authors suggested that this could indicate POB 39040 that the outer peripheral wall of enclosure C was built Ramat Aviv during the backfilling of Enclosure D but admit that Tel Aviv 69978 this may be a “far-reaching” conclusion. The date avail- Israel able for enclosure A (8617–8315 cal. BC. Charcoal. E-mail: [email protected] Sample extracted from beneath a fallen pillar fragment) may represent the end of the enclosure’s use-life and it is significantly younger than the dates available for Enclosures C and D. These dates and considering the References “square-like” layout of enclosure A, allowed the authors to conclude that Enclosure A seems generally younger Abbo, S. & Gopher, A., 2017. Near Eastern plant domesti- than enclosures C and D’ (O. Dietrich et al. 2013, 41). cation: a history of thought. Trends in Plant Science 2. Notably, according to L. Dietrich et al. (2019), the earl- 22(6), 491–511. ier phases of the enclosures are attested by the exten- Abbo, S., S. Lev-Yadun, & A. Gopher, 2010. Agricultural sive re-use of pillars (bearing carved figures that are origins: centres and noncenter; a Near Eastern sometimes hidden by the peripheral walls). reappraisal. Critical Reviews in Plant Science 29(5), 3. As noted by Peters and Schmidt (2004, 209), the fox 317–28. assumes a special role and is ‘overrepresented’ in the Aurenche, O. & S.K. Kozlowski, 1999. La naissance du faunal remains: ‘Only for fox, a certain similarity Néolithique au Proche-Orient. Paris: Errance. between bone refuse and artistic representation can Banning, E.B., 2011. So fair a house: Göbekli Tepe and the be pointed out: In the refuse, fox remains are counted identification of temples in the Pre-Pottery Neolithic in a rather high frequency (n = 971, Table 1), even out- of the Near East. Current Anthropology 52(5), 619–60. numbering remains of wild boar and reaching the Benz, M. & Bauer, J., 2013. Symbols of power – symbols of amount of sheep/goats. This somewhat surprising crisis? A psycho-social approach to Early Neolithic result may be connected with the exploitation of its symbol systems. Neo-Lithics: The Newsletter of pelt and/or the utilisation of fox teeth for ornamental Southwest Asian Neolithic Research 2/13, 11–24. purposes.’ Beyond that point, based on the data avail- Bar-Yosef, O., 2014. Was Göbekli Tepe culture a chiefdom able at the time, they contemplate on the fox and say that failed? in Settlement, Survey and Stone. Essays on that ‘Additionally, a specific worship of foxes may be Near Eastern prehistory in honour of Gary Rollefson, reflected here, which in fact finds parallels in the num- eds B. Finlayson & C. Makarewicz. Berlin: Ex oriente, ber of depictions on the stone pillars.’ (Peters & 159–68. Schmidt 2004, 209). So the totemic view is not a simple Çelik, B., 2010. Hamzan Tepe in the light of new finds. equation and would take deeper insights. Documenta Praehistorica 37, 257–68. Çelik, B., 2011. Karahan Tepe: a new cultural centre in the Acknowledgements Urfa area in Turkey. Documenta Praehistorica 38, 241–54. Clare, L, O. Dietrich, J. Notroff & D. Sönmez, 2018. We thank Lee Clare and Oliver Dietrich for their permis- Establishing identities in the Proto-Neolithic: ‘history sion to use and reproduce the figures in this paper. We making’ at Göbekli Tepe from the late tenth millen- thank the reviewers for their constructive comments. nium cal BCE, in Religion, History, and Place in the

355

Origin of Settled Life, ed. I. Hodder. Louisville (CO): Approaching Monumentality in Archaeology, ed. Colorado University Press, 115–36. J.F. Osborne. (IEMA Proceedings Vol. 3.) Albany Dietrich, L., J. Meister, O. Dietrich, J. Notroff, J. Kiep, (NY): SUNY Press, 83–105. J. Heeb, A. Beuger & B. Schütt, 2019. Cereal process- Özdoğan, M., 1999. Northwestern Turkey: Neolithic cul- ing at Early Neolithic Göbekli Tepe, southeastern tures in between the Balkans and Anatolia, in Turkey. PLOS ONE 14(5), p.e0215214. Neolithic in Turkey. The cradle of civilization,eds Dietrich, O., 2017a. Enclosure A, a short overview. The Tepe M. Özdoğan & N. Basgelen. Istanbul: Arkeoloji ve Telegrams – News & Notes from the Göbekli Tepe Sanat Xayınları. Research Staff. https://tepetelegrams.wordpress. Özdoğan, M., 2014. The quest for new criteria in deﬁning com/2017/01/05/enclosure-a-a-short-overview/ the emergence and the dispersal of Neolithic way of Dietrich, O., 2017b. Enclosure B, a short overview. The Tepe life, in The Neolithic Transition in the Mediterranean, Telegrams – News & Notes from the Göbekli Tepe eds C. Manen, T. Perrin & J. Guilaine. Paris: Research Staff. https://tepetelegrams.wordpress. Errance, 77–89. com/2017/02/03/enclosure-b-a-short-overview/ Peters, J. & K. Schmidt, 2004. Animals in the symbolic Dietrich, O., Ç. Köksal-Schmidt, J. Notroff, K. Schmidt & world of Pre-Pottery Neolithic Göbekli Tepe, south- Ç. Kürkçüoğlu, 2012a. Göbekli Tepe: a Stone Age rit- eastern Turkey: a preliminary assessment. ual center in southeastern Turkey. Actual Archaeology Anthropozoologica 39(1), 179–218. Magazine Summer (2), 30–51. Peters, J., K. Schmidt, O. Dietrich, & N. Pöllath, 2014. Dietrich, O., Heun, M., Notroff, J., Schmidt, K. & Göbekli Tepe: agriculture and domestication, in Zarnkow, M., 2012b. The role of cult and feasting Encyclopedia of Global Archaeology, ed. C. Smith. in the emergence of Neolithic communities. New evi- New York (NY): Springer, 3065–8. dence from Göbekli Tepe, south-eastern Turkey. Piesker, K., 2014. Göbekli Tepe. Bauforschung in den Antiquity 86, 674–95. Anlagen C und E in den Jahren 2010–2012. Dietrich, O., Ç. Köksal-Schmidt, J. Notroff & K. Schmidt, Zeitschrift für Orient-Archäologie 7, 14–54. 2013. Establishing a radiocarbon sequence for Schmidt, K., 2000. Göbekli Tepe, southeastern Turkey: a Göbekli Tepe. State of research and new data. preliminary report on the 1995–1999 excavations. Neo-Lithics 1(13), 36–41. Paléorient 26(1), 45–54. Dietrich, O., Ç. Köksal-Schmidt, J. Notroff & K. Schmidt, Schmidt, K., 2001. Göbekli Tepe and the early Neolithic 2014. Göbekli Tepe. Preliminary report on the 2012 sites of the Urfa region: a synopsis of new results and 2013 excavation seasons. Neo-Lithics 1, 11–17. and current views. Neo-Lithics 1(1), 9–11. Gopher, A., S. Lev-Yadun & S. Abbo, 2017. Domesticating Schmidt, K., 2002a. Göbekli Tepe: southeastern Turkey, the plants in the Near East, in Quaternary of the Levant: seventh campaign, 2001. Neo-Lithics 1(2), 23–5. Environments, climate change and humans, eds Schmidt, K., 2002b. The 2002 excavations at Göbekli Tepe Y. Enzel & O. Bar-Yosef. Cambridge: Cambridge (southeastern Turkey): impressions from an enig- University Press, 737–42. matic site. Neo-Lithics 2(2), 8–13. Güler, G., B. Çelik & M. Güler, 2013. New Pre-Pottery Schmidt, K., 2003. The 2003 campaign at Göbekli Tepe Neolithic sites and cult centres in the Urfa region. (southeastern Turkey). Neo-Lithics 2(3), 3–8. Documenta Praehistorica 40, 291–303. Schmidt, K., 2006. Animals and a headless man at Göbekli Haklay, G. & A. Gopher, 2015. A new look at shelter 131/ Tepe. Neo-Lithics 2(6), 38–40. 51 in the Natuﬁan site of Eynan (Ain-Mallaha), Schmidt, K., 2007. Göbekli Tepe excavations 2005. Kazı Israel. PloS ONE 10(7), e0130121. sonuçları toplantısı 2 cilt,97–110. Haklay, G. & Gopher, A., 2019. Architectural planning and Schmidt, K., 2008. Göbekli Tepe – enclosure C. Neo-Lithics measuring in the Pre-Pottery Neolithic site of 2(8), 27–32. Çayönü, Turkey. Paléorient 45(1), 7–17. Schmidt, K., 2010. Göbekli Tepe – The Stone Age sanctuaries. Hauptmann, H., 1993. Ein Kultgebäude in Nevali Çori, New results of ongoing excavations with a special in Between the Rivers and Over the Mountains: focus on sculptures and high reliefs. Documenta Archaeologica Anatolica et Mesopotamica Alba Palmieri Praehistorica 37, 239–56. dedicata, eds M. Frangipane, H. Hauptmann, Schmidt, K., 2012. Göbekli Tepe: A Stone Age sanctuary in M. Liverani, P. Mathhiae & M. Mellink. Rome: south-eastern Anatolia. Berlin: Ex Oriente. Universita de Roma ‘La Sapienza’,37–69. Watkins, T., 2004. Building houses, framing concepts, con- Helmer, D., L. Gourichon & D. Stordeur, 2004. À l’aube de la structing worlds. Paléorient 30(1), 5–23. domestication animale. Imaginaire et symbolisme Watkins, T., 2010. New light on Neolithic revolution in animal dans les premières sociétés néolithiques du south-west Asia. Antiquity 84, 621–34. nord du Proche-Orient. Anthropozoologica 39(1), 143–63. Lev-Yadun, S., A. Gopher & S. Abbo, 2000. The cradle of agriculture. Science 288(5471), 1602–3. Author biographies Notroff, J., O. Dietrich & K. Schmidt, 2014. Building monu- ments – creating communities. Early monumental Gil Haklay is an archaeologist at the Israel Antiquities architecture at Pre-Pottery Neolithic Göbekli Tepe. Authority, Archaeological Research Department, Prehistoric

356

Branch. He graduated in Architecture at ‘La Sapienza’ Avi Gopher (PhD 1985, Hebrew University) is Professor of University of Rome, and in Archaeology at Tel Aviv Archaeology in the Institute of Archaeology, Tel Aviv University (where he is currently a PhD candidate). His University. His major research is focused on the Neolithic research is focused on prehistoric architecture and ground period and the archaeology of plant domestication in the stone tools. He has also directed excavations at the Pre- Near East. He also co-directs excavations and research on Pottery Neolithic A site of Bir el-Maksur, Israel. the late Lower Palaeolithic period at Qesem Cave, Israel.

357